Re: [PATCH v2] vduse: Validate vq_num in vduse_validate_config()

[Michael S. Tsirkin]

On Mon, Nov 28, 2022 at 12:13:10PM +0100, Stefano Garzarella wrote:
> On Mon, Nov 28, 2022 at 01:58:00PM +0300, Dan Carpenter wrote:
> > On Mon, Nov 28, 2022 at 11:53:12AM +0100, Stefano Garzarella wrote:
> > > On Mon, Nov 28, 2022 at 12:36:26AM -0800, Harshit Mogalapalli wrote:
> > > > Add a limit to 'config->vq_num' which is user controlled data which
> > > > comes from an vduse_ioctl to prevent large memory allocations.
> > > >
> > > > This is found using static analysis with smatch.
> > > >
> > > > Suggested-by: Michael S. Tsirkin <mst@xxxxxxxxxx>
> > > > Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@xxxxxxxxxx>
> > > > ---
> > > > v1->v2: Change title of the commit and description, add a limit to
> > > > 	vq_num.
> > > >
> > > > Note: I think here 0xffff is the max size of vring =  no: of vqueues.
> > > > Only compile and boot tested.
> > > > ---
> > > > drivers/vdpa/vdpa_user/vduse_dev.c | 3 +++
> > > > 1 file changed, 3 insertions(+)
> > > >
> > > > diff --git a/drivers/vdpa/vdpa_user/vduse_dev.c b/drivers/vdpa/vdpa_user/vduse_dev.c
> > > > index 35dceee3ed56..31017ebc4d7c 100644
> > > > --- a/drivers/vdpa/vdpa_user/vduse_dev.c
> > > > +++ b/drivers/vdpa/vdpa_user/vduse_dev.c
> > > > @@ -1440,6 +1440,9 @@ static bool vduse_validate_config(struct vduse_dev_config *config)
> > > > 	if (config->config_size > PAGE_SIZE)
> > > > 		return false;
> > > >
> > > > +	if (config->vq_num > 0xffff)
> > > 
> > > What about using U16_MAX here?
> > 
> > Where is the ->vq_num stored in a u16?  I looked for this but didn't
> > see it.
> 
> I thought vq_num referred to the number of elements in the vq (like
> .get_vq_num_max), since this patch wants to limit to 0xffff.
> 
> But it actually refers to the number of virtqueue, so @Harshit why do we
> limit it to 0xffff?
> 
> Maybe we should explain it in the commit message or in a comment.
> 
> Thanks,
> Stefano

This limit is somewhat arbitrary.
However, currently virtio pci and ccw are limited to a 16 bit vq number.
While MMIO isn't it is also isn't used with lots of VQs due to
current lack of support for per-vq interrupts.
Thus, the 0xffff limit on number of VQs corresponding
to a 16-bit VQ number seems sufficient for now.

Feel free to put the above in a code comment.

-- 
MST